Sorry @samj, but I’m still confused.
The title of this thread says “The OSAID requires training data to be shared”, but the current FAQ says it allows unshareable data.
Which of the two declaration is false?
Given @nick clarified that the negative votes from the Llama team were are error and “should be converted to 0” and you have shown that this lead to a definition that requires both training and testing data, so, assuming good faith, the FAQ should be fixed before the RC1, shouldn’t it?
PS: I agree that as long as the training and testing data are available to the public under the same terms that allowed their usage from the builders in the first place, we can still count the AI system as “Open Source AI”, since the 4 freedoms are still granted even if the builders cannot directly distribute them.
The proposal that @nick closed a few week ago went in that direction.
Maybe it could be reopened and discussed now?